Datasheet

MCP4706/4716/4726
DS22272C-page 60 © 2011-2012 Microchip Technology Inc.
7.7 Integral Nonlinearity (INL)
The Integral Nonlinearity (INL) error is the maximum
deviation of an actual transfer function from an ideal
transfer function (straight line).
In the MCP47X6, INL is calculated using two end points
(zero and full scale). INL can be expressed as a per-
centage of full scale range (FSR) or in a fraction of an
LSb. INL is also called relative accuracy. Equation 7-4
shows how to calculate the INL error in LSb and
Figure 7-2 shows an example of INL accuracy.
EQUATION 7-4: INL ERROR
FIGURE 7-2: INL Accuracy Example.
7.8 Differential Nonlinearity (DNL)
The Differential Nonlinearity (DNL) error (see Figure 7-3)
is the measure of step size between codes in actual
transfer function. The ideal step size between codes is
1 LSb. A DNL error of zero would imply that every code
is exactly 1 LSb wide. If the DNL error is less than 1 LSb,
the DAC ensures monotonic output and no missing
codes. The DNL error between any two adjacent codes
is calculated as follows:
EQUATION 7-5: DNL ERROR
FIGURE 7-3: DNL Accuracy Example.
INL
V
OUT
V
Ideal
–()
LSb
---------------------------------------=
Where:
INL is expressed in LSb.
V
Ideal
= Code*LSb
V
OUT
= The output voltage measured with
a given DAC input code
010001000
Analog
Output
(LSb)
DAC Input Code
011 111100 101
1
2
3
4
5
6
0
7
110
Ideal Transfer Function
Actual Transfer Function
INL = < -1 LSb
INL = 0.5 LSb
INL = - 1 LSb
DNL
ΔV
OUT
LSb–
LSb
----------------------------------=
Where:
DNL is expressed in LSb.
Δ
V
OUT
= The measured DAC output
voltage difference between two
adjacent input codes
010001000
Analog
Output
(LSb)
DAC Input Code
011 111100 101
1
2
3
4
5
6
0
7
DNL = 2 LSb
DNL = 0.5 LSb
110
Ideal Transfer Function
Actual Transfer Function